With all three stuck, there was no hope, so I added a plastic piece for the hole.

Here's the connecting pin. It had to be bent creatively because the hole didn't end up where it should have. Hot glue makes a great temporary mount.

So after all that servo action, it turns out this approach wasn't acceptable. The linear servo didn't behave as well installed as it did free. Luckily the original design was for a standard rotary servo. All I needed to do was make some new holes and standoffs. And the connecting pin needed to be a little longer (although not quite this long; that's for ease of assembly).

Here's the whole thing mostly ready to go. The main demo functionality is ready. What's missing is the auto-home feature to find the fake boot sector on power-up, and the read/write LED on the head. The LED is on the breadboard, though. Beyond that, it's etching a printed circuit board and fabricating an enclosure. This thing is definitely not portable, although I will be lugging it to class.

The calibration marks indicate my attempts at determining how many tracks and sectors to define. Obviously, capacity isn't the concern; it's making this interesting and informative. To this end, I went with six tracks and 20 sectors. the tracks are marked from center to 6 o'clock. The other marks are cluster tests. I'll wash all this off later and put in nice clusters.

This video is an example of it in action. The test isn't synchronized to the marks, but the basic idea is here.

And a closeup of the display. The LED is the read/write indicator.